Vehicle detection methods from an unmanned aerial vehicle platform

Abstract

Vehicle detection is one of the key requirements for traffic surveillance. In most Intelligent Transportation Systems(ITS), cameras are installed in fixed places, which limits the field of view(FOV) of the cameras. This paper presents a new vehicle detection approach by analysing airborne video captured from a quad rotor unmanned aerial vehicle(UAV). Different detection methods on videos of moving and static vehicles are used to meet the requirements of traffic surveillance. Moving vehicles are detected by a feature point tracking method based on the combination of scale invariant feature transform(SIFT) and Kanada-Lucas-Tomasi(KLT) matching algorithm, and an effective clustering method, while static vehicles are recognized by analysing the blob information after automatic road extraction. In order to increase the precision of detection, some pre-processing methods are added into the surveillance system. Experimental results indicate that the proposed approaches of vehicle detection can be realized with a high identification ratio.